Is being named by ChatGPT cheaper than paying for Google Ads?

You have been paying Google for clicks for years. The invoice arrives monthly, the leads arrive in roughly the numbers you expect, and when you pause the campaign the phone goes quiet. It works, in the way a taxi meter works. Now someone on your team has pointed out that buyers are asking ChatGPT and Gemini who to hire, and that the answer does not mention you.

The founder's question is financial. If we spent the same money on being named by AI assistants instead of buying clicks, would we win more clients, fewer or the same, and how long would it take?

The ground is already moving under paid search. Seer Interactive's November 2025 study of 42 organisations found that paid click-through on searches showing an AI Overview fell 68% between June 2024 and September 2025, from 19.70% to 6.34%. The same study found that brands cited inside the AI Overview earned 91% higher paid click-through than brands that were not. The two channels are partners with very different cost curves, and this guide sets out how to split a finite budget between them.

#How does paid search really cost you money?

Paid search charges you for attention, one click at a time, for as long as you run it. Google auctions the click, you win it, and a stranger lands on your page, most of whom will not become clients. Three things follow:

• The cost per client is set by other people. If two competitors bid harder, your price rises without your service changing.
• Nothing is left behind. A click that does not convert has no residual value.
• The channel stops the day you stop, which is why founders describe paid search as a treadmill.

None of that makes it a bad channel: for enquiries next week, it is often the only one that delivers. It is an operating cost, and should be judged as one.

#How do I work out cost per client from Google Ads?

Divide what you spent by the clients it produced, not by the leads. Cost per client is the only figure that compares fairly with any other channel.

#Step 1: Pull twelve months of spend and clicks

Take total spend and clicks from Google Ads for the last twelve months, so seasonal swings even out.

#Step 2: Count the conversion chain

Record how many clicks became enquiries, how many enquiries became first calls and how many calls became clients, using Google Ads conversion tracking and your CRM.

#Step 3: Multiply the leaks together

For example, suppose 1 in 20 clicks becomes an enquiry, 1 in 2 enquiries becomes a call, and 1 in 4 calls becomes a client. You then pay for 20 x 2 x 4 = 160 clicks for every client won, so your cost per client is 160 times your average cost per click. The rates here are illustrative; use your own.

#Step 4: Check the price of the terms you rely on

Google's Keyword Planner, free inside a Google Ads account, shows estimated monthly searches and what it typically costs for your ad to show. Set the location to the United Kingdom, look up the three terms that bring you most clients, note the date and repeat each quarter.

#What is the average cost per click for B2B keywords in the UK?

There is no single figure worth quoting. Prices vary widely by term and sector and change with the auction, and we have found no dated, published UK benchmark for B2B terms we would stand behind. The figure that matters is your own: what you pay for the terms that bring clients, and how many clicks it takes to win one.

#How do Google Ads, LinkedIn Ads, ChatGPT ads and being named compare?

The three paid channels charge you to be seen; being named in an AI answer is earned. The table sets out how each works for a UK B2B firm today.

The UK is not among the twelve countries where Google sells ads inside AI Overviews, and OpenAI's ads page says the Plus, Pro, Business, Enterprise and Education plans show no ads, so a buyer on a company account never sees one.

#What does being named by an AI assistant cost?

The cost of being named by ChatGPT, Claude, Gemini or Perplexity is the cost of becoming the obvious answer, paid mostly up front, and then very little per recommendation after that.

When a buyer asks an assistant "which firms handle payroll for construction companies in the Midlands", the answer is not sold by auction. The assistant draws on your website, the coverage and directories that mention you, your reviews and case studies, and how consistently they agree. What it cites is overwhelmingly earned: Muck Rack's May 2026 analysis of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% were earned media and 0.3% were paid or advertorial. There is no invoice for the mention. ChatGPT now sells ads beside the answer, but OpenAI says advertising does not influence the answers ChatGPT gives.

So the spend is different in kind: a website that says plainly what you do and for whom, in a form AI crawlers can read; articles and case studies that answer buyers' questions; coverage that repeats the same facts about you; and the time of senior people to put their names to it. That looks expensive next to a month of clicks, and in the first year it often is. The difference is month thirteen, when last spring's article, the trade press piece and the case study are all still there, and you are not paying again for each recommendation they produce. How assistants choose which firms to name is explained in how ChatGPT decides which firms to recommend.

What outside help costs against a pay-per-click budget depends on who does the work: UK market ranges are in how much AI search optimisation costs in the UK, and our prices are on what it costs. A programme is a fixed cost that builds assets; paid search is a variable cost that buys clicks.

#Why do the two cost curves run in opposite directions?

In our view, the cost of an enquiry from paid search is flat at best and rising at worst, while the cost of an enquiry from being named starts high and falls.

Picture two lines on a chart. The paid line is roughly horizontal: you pay about the same for the hundredth client as for the first, and if the auction heats up, or fewer people click ads because an AI answer sits above them, it tilts upwards. Better pages and tighter keywords help, but the unit is still a click bought at an auction that resets every day. The earned line starts well above it, because you are building assets and getting little back for some months, then bends down as each new piece adds to the evidence, being named brings clients, and clients become case studies. What locks in and what does not is set out in whether you can afford to wait a year on AI search.

Where the lines cross depends on your sector, your starting point and how consistently you do the work. We have no dataset across UK B2B firms that supports a typical crossover month, so we do not quote one.

#Should I advertise on ChatGPT instead of Google Ads?

Not instead, and not yet as a main channel for most B2B firms. ChatGPT ads have been live for UK users on the Free and Go plans since 11 August 2026, and OpenAI says self-service buying is launching across Europe. They reach people at the moment they ask a question, but only on the plans that carry ads, and they sit beside the answer rather than in it. If your buyers use personal accounts, run a small test judged on cost per client; if they use company accounts, most will never see the ad.

#Which channel brings the better customer?

An enquiry that arrives because an assistant named you tends to be further along and less price-driven than one that arrives from an advert, because the buyer has already been told you are a credible option.

A paid click catches a buyer comparing four suppliers in four tabs, and they arrive cold. An AI recommendation usually comes when the buyer is building a shortlist, and shortlists decide a great deal: the 6sense 2025 Buyer Experience Report found that buying groups had ranked their preferred vendors before first contact in 94% of cases and bought from that first choice 77% of the time.

Helm Club, a founder membership community, shows what the channel can become: "1 in 5 new members this year came through an AI recommendation." The full story is in the Helm Club case study.

We have seen no study measuring the uplift in deal size across UK B2B firms, so we claim none. More on buyers who arrive pre-sold is in what your website should do now visitors arrive pre-sold by AI.

#When does AI search break even against paid search?

When your cumulative spend on being named, divided by the clients it has brought, falls below your paid cost per client. No general month can be quoted, so measure it for your own firm and plan for a gap of some months in which paid search carries the business.

ID-Pal gives a dated example of how being named builds: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." Assistants that search the web live can reflect new pages within weeks; answers from a model's memory change only when it is retrained, which takes months.

The gap is a real cash cost, and a board that has not been told to expect it will lose its nerve. So stage the shift. For example, a firm might plan:

These proportions are illustrative; what matters is deciding them in advance and reviewing them each quarter. To put that to a board, see how to get your board to approve budget for AI search, and for what doing nothing costs, see what it is costing you not to be recommended.

#Why should we keep advertising anyway?

Do not switch paid search off; keep it as a baseline while you build the asset, for four reasons.

First, the channels help each other: in Seer Interactive's data, brands cited in an AI Overview earned 91% higher paid click-through (7.89% against 4.14%). Second, speed: paid search is the only lever that produces enquiries next week. Third, coverage: some buyers still click adverts, even as AI answers take more of the page, as we explain in why ranking first on Google does not guarantee a ChatGPT mention. Fourth, diversification: two channels are safer than one.

#What does the comparison look like for a smaller firm?

The logic holds at a smaller scale. A firm with a modest budget can still fix its website so it says plainly what it does and for whom, and publish one useful answer to a buyer question each month under the founder's name. In a region or a niche, that is often enough to start being named.

#What to do this month

1. Work out your real cost per client from twelve months of paid search data, using the four steps above.
2. Look up your three most valuable terms in Keyword Planner, set to the United Kingdom, and record the figures with the date.
3. Add "an AI assistant such as ChatGPT" to the "how did you hear about us" options on every enquiry form, and ask the question on every first call.
4. Decide, in writing, what share of next year's budget moves from clicks to building the asset, and what evidence would make you move more.
5. Pick the first three buyer questions your firm should be the answer to, and give each to a named person to write.

#Frequently asked questions

#Can we just keep running Google Ads and ignore AI assistants?

You can, and for a while it will work. The risk is that more buyers form their shortlist in an assistant, and you find out late that a competitor has become the name it gives. Treating paid search as the whole plan is a bet that buyers keep clicking ads as AI answers take more of the page.

#What share of our budget should move across in the first year?

Enough to build the asset properly, not so much that enquiries collapse first: for example, around a fifth in year one, with more once AI-sourced enquiries show. Decide it with the board.

#How do we know if AI-sourced enquiries are actually arriving?

Ask, and check. Add the option to your enquiry form and ask on every first call. Separately, run a fixed list of 20 to 50 buyer questions through the assistants several times a month, counting a win only when it repeats. Helm Club's results are reported the same way: "Qualified calls a week, reported on one page every Monday and checked against Helm's own member records rather than a platform dashboard." Our guide to tracking whether ChatGPT recommends your firm covers the set-up.

#Does this apply to a firm selling a low-value service?

Less strongly. AI recommendations matter most where a buyer chooses a supplier carefully, the purchase is significant and the decision is professional. For a cheap, quick purchase, paid search and price still do most of the work. For most founder-led B2B firms with revenue of £1m to £20m+, the purchase is considered, and the argument applies.

#We are profitable on paid search today. Why change anything?

Because that profit depends on paying the auction for as long as you want the enquiries, and on buyers continuing to click ads. The asset you build to be named keeps producing after the spend stops. You can keep the paid channel running while you build it.

In our view, adverts are usually cheaper this quarter and being named is usually cheaper over any period that matters to a founder. If you want to know where you stand before you move any budget, book a free Growth Review: 30 minutes with David Adams, and three fixes in writing afterwards.

#Sources